448 



SCIENCE. 



[N. S. Vol. III. No. 64. 



characters show, however, that it is in no way 

 closely related to C. alpinus, but, on the con- 

 trary, is a special offshoot from some South 

 American dog of the ordinary type. Procyon 

 nrsinus, while showing certain characters which 

 prove it to stand nearer the ancestral stock 

 than do the existing species, is considerably 

 larger than any of the latter. 



Turning now to the living Carnivora, Mr. 

 Winge gives very extended and elaborate dis- 

 cussions of the specific characters and of the 

 individual variation in both size and color of 

 most of the forms represented in the collection. 

 While the author's tendency to reduce the num- 

 ber of species to the minimum must detract 

 from the critical value of this part of the work, 

 the facts recorded will be of the utmost use to 

 all workers on South American mammals. The 

 discussions of the species of Felis and Canis are 

 especially important. 



For the most part the bones found in the 

 caves agree perfectly with those of the living 

 representatives of the variovis species. There 

 are, however, a few exceptions to this rule. 

 Thus, only a few of the cave remains of Felis 

 onca are of the same size as those of the ordin- 

 ary existing Jaguar. Most of them represent 

 animals which were about the size of F. tigris. 

 There is scarcely any difference in the teeth, 

 but in the bones the discrepancy is very notice- 

 able. Although the cave Jaguars average 

 much larger than those now living, one of the 

 latter occasionally fully equals the largest of the 

 former. The single perfect skull of Icticyon 

 venaticus from Lapa dos Tatus differs remark- 

 ably from recent skulls of the same species. 

 The nasal bones are much more produced both 

 before and behind, while the whole skull is 

 larger ; the rostrum is considerably broader in 

 proportion to the brain case, and the zygomatic 

 arches are more flaring posteriorly. If the dif- 

 ferences between this skull and that of the re- 

 cent specimen as figured on plate V. are in no 

 way due to age and sex, few mammalogists 

 would hesitate to separate the animals specifi- 

 cally. Mr. Winge, however, does not consider 

 such a course advisable, though he admits that 

 the Lapa dos Tatus skull may represent a ' geo- 

 logical race ' (' men maaske er det en geologisk 

 race '). 



As the result of his studies of the interrela- 

 tionships of the Carnivora in general, Mr. 

 Winge gives the following table of super-generic 

 groups (p. 46, 47). 



Carnivora primitiva. 

 Hysenodontidse. 



Provlverrinl. 

 Mesonychini. 

 Hysenodontini. 

 Arctocyonidae. 

 Carnivora vera. 



Herpestoidei. 



Amphictidse. 



Palseonictidse. 



Felidse. 



Felini. 



Machserodontini. 

 Viverridse. 



Viverrini. 



tim. 

 Hysenidae. 

 Arctoidei. 



Ursidae. 



Canini. 



XJrsini. 

 Procyonidse. 

 Mustelidse. 



Mustelini. 



Melini. 



Lutrini. 

 Otariidse. 



Trichechini. 



Otariini. 

 Phocidse. 



This arrangement differs in many details 

 from that recently adopted by Flower and 

 Lydekker.* The latter authors divide the 

 order into fifteen families and one hundred and 

 six genera, while Winge recognizes the same 

 number of genera and only twelve families. Per- 

 haps the most noticeable peculiarity of the pres- 

 ent classification is the treatment of the bears, 

 dogs and raccoons. The two former, or the fam- 

 ilies Ursidx and Canidx of Flower and Lydekker 

 and of Zittel,f are here treated as subfamilies 

 of the family Ursidse, while the Frocyonidx are 

 kept distinct. Canis and Icticyon are thus 

 brought close to Ursus, while Procyon is placed 

 in a different group. 



* Mammals Living and Extinct, 1891. 



t Handb. der Palaentologie, Mammalia, 1892-1893. 



